This is a report on the second effort to formulate an effective question for determining the ethnicity of respondents to Great Britain's 1981 census. Problems encountered in the first pilot test and the two alternate survey forms tried in the second test are described. Included in the present version are additional questions on education, employment, language spoken at home, car ownership, and household accommodations. The response to the survey, which included 2,000 households, is reviewed and the effectiveness and error in response are discussed for individual questions. Also discussed are questions that respondents did not like. Appended to the report are a description of sampling procedures and field methodology and copies of both forms used in the survey. (GC)